Ripple voltage

iHearAll

iHearAll

Well-Known Member
Instead of only using a capacitor bank on the dc side of a rectifier to reduce the ripple, can I use an a bank of Ac caps in parallel to the rectifier as well?
I have some 450vac 300uF caps on hand I would use. I may have only 3-4 of them in a box in the closet.

I would imagine that the phase lag would already reduce the Vp- p ripple but maybe my understanding of the phase lag is skewed.

Im away from my bench right now and will just test this later but if someone knows off hand before then, I am intertested to hear.
Thanks
 
iHearAll

iHearAll

Well-Known Member
So, according to my math, to run 5 count clu048-1212 in series @170vdc I would around 3mF of capacitance on the dc side of my rectifier. Would I benefit from AC caps on the Ac side of the rectifier?
1495918952643587157238.jpg
 
CannaBruh

CannaBruh

Well-Known Member
You could do a 3terminal regulator and then heavily filter that, or even add some hf filtering after your ccc, but with good regulation and some filtering I'm not sure what you'd still be chasing? For caps on the primary side would be more for emissions and not sure you want to tackle your output ripple here, but maybe there's some tricks I'm not aware of.

nevermind I see your output is some 120V and some current that will be high enough to drive some what looks like COBS or something?

Why rectify it with diodes? Could you not build your rectifier out of your LEDs?
 
iHearAll

iHearAll

Well-Known Member
CannaBruh said:
You could do a 3terminal regulator and then heavily filter that, or even add some hf filtering after your ccc, but with good regulation and some filtering I'm not sure what you'd still be chasing? For caps on the primary side would be more for emissions and not sure you want to tackle your output ripple here, but maybe there's some tricks I'm not aware of.

nevermind I see your output is some 120V and some current that will be high enough to drive some what looks like COBS or something?

Why rectify it with diodes? Could you not build your rectifier out of your LEDs?
Click to expand...
Yea I was planning in using 5 cobs each at 35v 1a. So, an easy 170vdc string.
I have done the led rectifier you mention but it was similar to a strobe light. Anyway, a ballast isn't expensive for a situation like this, maybe 45$ then some shipping. But If I could manage to just store the lost energy to then absorb with the cobs i'd be pretty damn efficient and probably only 15-20$ aftrt shipping
 
Last edited:
nfhiggs

nfhiggs

Well-Known Member
iHearAll said:
Instead of only using a capacitor bank on the dc side of a rectifier to reduce the ripple, can I use an a bank of Ac caps in parallel to the rectifier as well?
I have some 450vac 300uF caps on hand I would use. I may have only 3-4 of them in a box in the closet.

I would imagine that the phase lag would already reduce the Vp- p ripple but maybe my understanding of the phase lag is skewed.

Im away from my bench right now and will just test this later but if someone knows off hand before then, I am intertested to hear.
Thanks
Click to expand...
Phase lag is irrelevant here. It only applies to inductive/capacitive loads like motors. Do not put caps on the AC side of your rectifier bridge. Put inductors between each filter cap in series with the load - what's commonly known as a "pi filter" :
https://en.wikipedia.org/wiki/Capacitor-input_filter

But all that aside, the real issue here is how are you going to control the current?
 
Last edited:
iHearAll

iHearAll

Well-Known Member
Capture+_2017-05-27-21-56-16.png
nfhiggs said:
Phase lag is irrelevant here. It only applies to inductive/capacitive loads like motors. Do not put caps on the AC side of your rectifier bridge. Put inductors between each filter cap in series with the load - what's commonly known as a "pi filter" :
https://en.wikipedia.org/wiki/Capacitor-input_filter

But all that aside, the real issue here is how are you going to control the current?
Click to expand...
It should self regulate by the voltage of the Ac maxing at 170 . so since the diodes work on a liner path upwards in voltage and current the diode will stay at 170 and less since I can assume I won't get a perfect 0v ripple in the end. Ill fuse the circuit in case of some noise some days or put a choke inductor on the DC. Thanks for explaining that tid about phase lag. That was mostly what I'm concerned with. It didnt make sense in that it would help the ripple since voltage and current were separated by time.

Edit: its not linear, idk why I said that. It has a consent slope upward, or almost constant. Predictable in the least.
 
Last edited:
nfhiggs

nfhiggs

Well-Known Member
iHearAll said:
View attachment 3950238
It should self regulate by the voltage of the Ac maxing at 170 . so since the diodes work on a liner path upwards in voltage and current the diode will stay at 170 and less since I can assume I won't get a perfect 0v ripple in the end. Ill fuse the circuit in case of some noise some days or put a choke inductor on the DC. Thanks for explaining that tid about phase lag. That was mostly what I'm concerned with. It didnt make sense in that it would help the ripple since voltage and current were separated by time.

Edit: its not linear, idk why I said that. It has a consent slope upward, or almost constant. Predictable in the least.
Click to expand...
34V per COB... I would do some testing to see what the current draw on your COB's are at 34V, using a variable power supply. The COBs can vary significantly from the ideal curves given in the data sheets. Another thing to consider is line voltage stability. AC mains in the US can vary from 108V to 132V. Just a few volts on the high side can have you popping COBs like popcorn. Also there is the possibility of thermal runaway. As COB's heat up, their forward voltage drops. If you have a constant voltage on them, then the current increases, which causes further heating, which causes their forward voltage to drop more. Your power supply has literally unlimited current. You need some method of limiting current.
 
iHearAll

iHearAll

Well-Known Member
nfhiggs said:
34V per COB... I would do some testing to see what the current draw on your COB's are at 34V, using a variable power supply. The COBs can vary significantly from the ideal curves given in the data sheets. Another thing to consider is line voltage stability. AC mains in the US can vary from 108V to 132V. Just a few volts on the high side can have you popping COBs like popcorn. Also there is the possibility of thermal runaway. As COB's heat up, their forward voltage drops. If you have a constant voltage on them, then the current increases, which causes further heating, which causes their forward voltage to drop more. Your power supply has literally unlimited current. You need some method of limiting current.
Click to expand...
The heatsink would keep the case temp down and if the case temp is down the cob is going to act like a power resistor on the line. So it would resemble screwing in a incandescent filament bulb on the line. If I bring my vdc to only 160-165v using only dc caps on the bridge rectifier and the temperature is accounted for, they'll sit happy. At that point i just need a 2a fuse in case of weather and brutal noise.

The cobs I am working with are citizen brand, so what i get would be within a quality control tolerance I can probably find on the data sheet. I would think I can use my variable dc supply to measure each and plot a characteristic curve for each individual cob too but 35vdc, although obtainable with my psu, is out of its purposed range. Its a 30vdc 10a vsu. But, it will go that high is it is only using a small current.

Anyway, I have some 470uF dc caps and fast blow fuses on hand as well, ill do some tests when I get to my bench with the chips.

Thnks!
 
iHearAll

iHearAll

Well-Known Member
My current DIY panel runs at 90vdc that I obtained with a series AC capacitor @ 100uF. I then rectified it and put my cheap Chinese cobs on the output side. No ripple smoothing and the current is self limited to 1.2 A by the chips. Its been running for a year or more now. Different design obviously
 
CannaBruh

CannaBruh

Well-Known Member
Why do you put a cap in series?

Most people are using constant current, so they don't run into "what happens when my load demands more when my line voltage is high today"

How important is ripple with these COBs? we use switching power supplies in audio circuits in studio use with minimal attention to ripple beyond brute force filtering and some hf filtering
 
iHearAll

iHearAll

Well-Known Member
CannaBruh said:
Why do you put a cap in series?

Most people are using constant current, so they don't run into "what happens when my load demands more when my line voltage is high today"

How important is ripple with these COBs? we use switching power supplies in audio circuits in studio use with minimal attention to ripple beyond brute force filtering and some hf filtering
Click to expand...
Ripple isn't to crucial it seems. There isn't much info on the use of cobs in such a manor since most engineers wouldn't bother doing the math to ballast at the component level. But since in a broke ass cheep ass I play around with ideas like this.

I used the frequency of the line @ 60hz to use some industrial capacitors as resistors with the Xc formula for reactance. Then with some trig I could figure out the overall impedence since the caps are going in the neg j direction. This light works like a charm bbtw and the heatsink are barely warm! 1495982992376-1394396234.jpg
Of course I put a fuse on the circuit as well and it has survived many lightning storms. Maybe my town has a pretty clean line signal?

Edit: in using this same design in a six cob circuit that which the cobs require a lower voltage. I think the pictured above circuit was for cobs in the 20-30v range. I use 16-25v range cobs in a six cob design and the same Xc reactance ballast design successfully. It too has been running over a year.
 
CannaBruh

CannaBruh

Well-Known Member
too early for j-notation :eyesmoke:

my maths say your XsubC is some 26.526 ohms (eng notation wants 3 digits of resolution no :lol: ocd bad habit of mine)

your circuit basically puts the onus of regulation on the power company, and in your experience has shown that you are able to get away with it at least in your area specific to your grid.

How was your resistive leg of your impedance triangle found do you use the sum of COB resistance found by forward voltage divide by forward current?
 
nfhiggs

nfhiggs

Well-Known Member
iHearAll said:
The heatsink would keep the case temp down and if the case temp is down the cob is going to act like a power resistor on the line. So it would resemble screwing in a incandescent filament bulb on the line. If I bring my vdc to only 160-165v using only dc caps on the bridge rectifier and the temperature is accounted for, they'll sit happy. At that point i just need a 2a fuse in case of weather and brutal noise.

The cobs I am working with are citizen brand, so what i get would be within a quality control tolerance I can probably find on the data sheet. I would think I can use my variable dc supply to measure each and plot a characteristic curve for each individual cob too but 35vdc, although obtainable with my psu, is out of its purposed range. Its a 30vdc 10a vsu. But, it will go that high is it is only using a small current.

Anyway, I have some 470uF dc caps and fast blow fuses on hand as well, ill do some tests when I get to my bench with the chips.

Thnks!
Click to expand...
I would suggest putting a 5W 1 ohm resistor in series with the COB string, for two purposes. 1, it will act as a current limiter, and 2, it will be easy to monitor your current with a $2 voltage display.
iHearAll said:
Ripple isn't to crucial it seems. There isn't much info on the use of cobs in such a manor since most engineers wouldn't bother doing the math to ballast at the component level. But since in a broke ass cheep ass I play around with ideas like this.

I used the frequency of the line @ 60hz to use some industrial capacitors as resistors with the Xc formula for reactance. Then with some trig I could figure out the overall impedence since the caps are going in the neg j direction. This light works like a charm bbtw and the heatsink are barely warm! View attachment 3950459
Of course I put a fuse on the circuit as well and it has survived many lightning storms. Maybe my town has a pretty clean line signal?

Edit: in using this same design in a six cob circuit that which the cobs require a lower voltage. I think the pictured above circuit was for cobs in the 20-30v range. I use 16-25v range cobs in a six cob design and the same Xc reactance ballast design successfully. It too has been running over a year.
Click to expand...
Interesting idea - basically using the cap as a current limiting resistor. As long as the cap is hefty enough to handle the current flow (and it obviously is), it'll work. Hella inefficient though - at 1 amp, that cap is burning 26W of energy off.
 
iHearAll

iHearAll

Well-Known Member
CannaBruh said:
huh capacitor can dissipate?
Click to expand...
Yea, in ac it looms like a resistor. Cheap ones can get hot. Ill check it later with a volt meter but the circuit looks like a voltage divider. So, the cap should have around 30v or so across it.
nfhiggs said:
I would suggest putting a 5W 1 ohm resistor in series with the COB string, for two purposes. 1, it will act as a current limiter, and 2, it will be easy to monitor your current with a $2 voltage display.

Interesting idea - basically using the cap as a current limiting resistor. As long as the cap is hefty enough to handle the current flow (and it obviously is), it'll work. Hella inefficient though - at 1 amp, that cap is burning 26W of energy off.
Click to expand...
Yea in using two 52uF in parallel t. Account for the current passing and theyr 400vac if I recall. They have an aluminum casing and run super cool though.
 
You must log in or register to reply here.
Top